by dcgillespie » Thu Jan 26, 2012 11:11 pm
Hi Bruce -- First, thanks to jgf for his helpful responses -- they are very helpful, and right on target.
For a better understanding, I too would encourage you to read my article (see my initial post of this thread for the site it is posted on) wherein I introduce the EFB modification as a very real and results oriented solution to the underlying performance concerns, as produced by the combination of bias system and transformer specifications used in the small Dynaco amplifiers. It is an easy read, designed for new comer and experienced folks alike, and explains what the problem is, why it occurs, and how the EFB modification is such a simple, effective solution.
To thoroughly understand the reasoning behind why the EFB modification is so effective with specifically the Dynaco amplifiers first requires an understanding of the use of load lines as drawn on a "family" of plate curves, and what the resulting performance is that a given load line represents. It also requires a little understanding of David Hafler, and what his design decisions were in the manufacturing of his output transformers. Finally, it also requires an understanding of the design compromises made in the small Dynaco amplifiers, and what the results of those compromises are.
As to your specific questions, let me give you some basic generalities that you might find helpful. First, let us assume a perfect power supply is in place, that provides a perfect required B+ voltage, of infinite current capability -- in other words, it is perfectly regulated. In it's most basic form then, the output stage represents the output transformer and the output tubes as two components wired electrically in series. This series combination of tube and transformer is then wired in parallel across the output of the power supply. From this it can then be seen, that the output tubes are the active elements that are transferring power from the power supply, through the primary winding of the output transformer, and on to the load connected to the secondary of the transformer.
With this connection, it can be shown that the maximum power transfer into the load will be had when the impedance of the load is made exactly equal to the impedance of the the generator -- or in this case, that of the tubes. That is why that the best performance is had when the load of the output transformer matches the optimum load the tubes require.
The optimum load that a tube requires varies with a whole host of issues. Factors such as operating mode (triode, pentode, or UL operation), bias scheme used (fixed or resistive cathode bias), and the voltage the screen grid operates at all have a huge impact in determining what is the optimum load is for a given set of tubes in a given setting.
For the case at hand, when traditional resistive cathode bias is used, because it is an additional resistive element in series with the tubes, it necessarily raises the impedance of the generator, and so a numerically higher load impedance is then required to maintain maximum power transfer. With traditional cathode bias, the ultimate power will always be lower because of the additional resistive element it represents, so it will always require a higher load impedance to produce the maximum power it is capable of.
On top of this factor, because of the significant AC influence on the screen grid with UL operation, it to reduces the maximum current flow that a tube is capable of flowing, so just as with the use of cathode bias, the load impedance must be raised when UL operation is used.
These effects are cumulative. The use of cathode bias raises the required load somewhat. UL operation raises the required load somewhat. UL operation with cathode bias raises the required load a lot.
The crux of the matter is that the Dynaco transformer was wound to produce maximum performance out of 6BQ5 tubes when operated in UL mode with fixed bias operation -- but their amplifiers were designed with cathode bias as a matter of economy. Therein lies the problem -- and the opportunity as spelled out in my article.
As a final comment to your question, while the use of the EFB modification produces a significant performance improvement in the Dyanco amplifiers, it's use in other designs that employ a more appropriate load for the operating conditions used (including bias system employed) will obviously not produce the size of performance increase as shown in the Dynaco amplifiers for the reasons explained. On the other hand, it can hardly hurt either. You may also be interested in a short followup article I wrote regarding the potential use of the EFB modification with Heath's #51-29 OPTs, as used in most of their 6BQ5 offerings. In that setting, it would produce a slight improvement in power output, but the impact on distortion is muted as those transformers are well designed for used with resistive cathode bias operation from the get-go. You can find that article (as well as the original one) under the "Dave's Lab" section of the Tronola website.
I think if you read both of these articles, it will help you gain a better understanding of what the modification does, and what it is capable of doing in other designs.
I hope this has helped!
Dave